The Pilot Worked. Nobody Uses It. Here's the Anatomy.
Pilots do not fail on technology. They fail on five structural omissions, all of which are fixable at kickoff for free.
The five omissions
No production owner — the vendor or the enthusiastic manager moves on and no name is attached to the running system. Median time to silent death is about four months.
It ran on personal infrastructure: someone's laptop, a personal API key, an individual account. The credential dies with the employee and often nobody notices for weeks.
Success criteria were never written, so "did it work?" becomes an opinion contest between the sponsor and the sceptics, which the sceptics win by default. It was piloted with the most enthusiastic user, who works around every defect and reports success while the median user hits the same defects and stops. And it lived beside the daily workflow rather than inside it — a new dashboard, a new URL, a new login, competing with the tools people already have open.
The design fix
Pilots ship into the tool people already open every morning: the CRM task list, the ticket queue, the phone, the inbox.
And every pilot gets a written kill date and kill criteria at kickoff, with kill as the default outcome in the absence of evidence.
Choose your pilot user badly on purpose
Run it with your most sceptical, busiest user rather than your keenest. The enthusiast generates false positives by compensating for defects.
Zombie pilots that neither live nor die consume the credibility budget for the next three projects.
List every AI or automation pilot from the last eighteen months and mark each one live, dead or zombie. The zombie count is what to fix before starting anything new.
Has run organisations above €100 million in revenue with teams exceeding 10,000 people.
Meet the team →Questions we get asked
Why do successful AI pilots never reach production?
Five structural reasons: no named production owner, personal infrastructure and credentials, unwritten success criteria, testing with an enthusiast rather than a sceptic, and living beside the daily workflow instead of inside it.
Who should test an automation pilot?
Your most sceptical and busiest user. Enthusiasts compensate for defects and generate false positives; the median user hits those same defects and quietly stops.
How long do abandoned pilots survive before anyone notices?
Roughly four months is the median time to silent death when no production owner is named — and often nobody notices at all until a credential expires.
Related insight
The AI Assistant That Does Everything Does Nothing
An agent's reliability collapses as you add systems it must be simultaneously correct about. The only agents that reach production…
AutomationYour Software Is Fine. The Gaps Between It Cost You Forty Hours a Week.
The recoverable waste in most operating companies sits in the transfers between systems, not inside any one of them. No vendor is …
AutomationAutomate a Broken Process and You Buy a Faster Broken Process
Automation makes a process permanent. Any step you encode without questioning becomes structurally harder to remove than it was th…
Where is your business still running manually?
We review your processes, identify the highest-impact automation opportunities and estimate their financial impact — before anything is built.